Alt Tag

FranOFran's Bot Templates

I basically use some customized versions of the Jed format. Credit: not required but appreciated a lot done so ⸜(。˃ ᵕ ˂ )⸝♡


Trying to make this guide ADHD and Dyslexia friendly! let me know if I can improve it's presentation!


Small Template

  • Strengths: Good for bots with a lot of info you need to be token efficient. I personally use this for multi bots
  • Weaknesses: Can make the JLLM misinterpret and mix traits making the character a little one dimensional and not do so well in different scenarios
  • Weaknesses: If you're using multi bots, you can remove the past (substantiates without direct mention unless) and put it in a direct prompt bracket (read more about after templates)
1
2
3
4
5
6
7
8
9
[CHARNAME;
- Looks=
- Demeanor= Outward Act
- Cognition= Mind related traits, Thinking processes
- Personality= Traits(cause/why trait exists, reinforcement of why such trait persists, how that trait affects whole personality)
- Past(substantiates without direct mention unless)=
- Abilities=
- Speech=
]

Medium Template

Strengths:

  • Good if you want your character to be more versatile in different scenarios and you're still getting used to having as much info per least tokens
  • Good method in between Small and large template

Weaknesses:

  • If you're using multi bots, you can remove the past (substantiates without direct mention unless) and put it in a direct prompt bracket (read more about after templates)
  • Speech examples really help if you're trying to make a canon character feel more accurate!
[Looks:
- Features=
- Outfit=
]

[Personality:

- Occupation=

- Reputation= how the character is outwardly perceived, publicly

- Demeanor = How others perceive the character in person/direct contact

- Self perception= How the character sees themselves

- Seeks being perceived= Optional

- Cognitive styles= mind, thought processes traits etc

- Interpersonal attitude= Describe how the character behaves towards others

- Intrapersonal attitude = Describe how CHARNAME behaves towards themselves internally

- Conflict reaction/Problem solving=

- Skills=

- About= info that doesn't fit under other categories like age etc

- Past(substantiates without direct mention unless)=

Communication style=
Speech style=Tone, pacing, range, voice descriptors, accent etc.
Speech patterns examples(reference only, avoid verbatim)= "your example here", "your example here", "your example here", "your example here"
]

Large Template

Strengths: Works well to make characters versatile across RP stages/scenarios

Weaknesses: Token heavy, might not be suitable if you're character has a lot of information you need to put in.

Notes:

  • If you're using multi bots, you can remove the past (substantiates without direct mention unless) and put it in a direct prompt bracket (read more about after templates)
  • Speech examples really help if you're trying to make a canon character feel more accurate!
[Looks:
- Body=height, skin, build etc
- Features=face stuff, remarkable things like scars etc
- Hair=
- Apparel=
- Outfit=
]

[Personality:
- Name=
- Skills=
- Occupation=
- Demeanor= 
- Reputation=
- Values=
- Beliefs=
- Habits=
- Mannerisms=

Big five OCEAN:
- Openness = openness to new experiences, creativity and intellectual curiosity.
- Consciousness= level of self-discipline, organization and goal-directed behaviour.
- Extraversion=  Interpersonal style, how outgoing, energetic, and sociable a person is.
- Agreeableness=Captures a persons tendency to be cooperative, compassionate, and kind.
- Neuroticism= Reflects emotional stability and the tendency to experience negative emotions. Stress tolerance, how adaptable they are to mental strain, stress response(fight, freeze, fight or fawn)

Emotional Response:
- Reactivity=How reactive your character is to outside forces, emotion or how easy they are impacted by these events.
- Regulation=How easy is for your character to come back from a state of mental strain, methods of doing it etc
- Triggers=Anything that causes an abrupt emotional or stress shift
- Conflict Resolution=
- Problem Solving=

Self Concept:
- Self Esteem=How confident they are as a whole
- Self Efficacy=how confident they are in their ability to accomplish/succeed their goals/tasks
- Self perception=how they see themselves
- Locus of control=Internal or external? maybe both depending on your character situation
- Seeks being perceived=how they want others to perceive them

Morals:
- Integrity= states how well someone upholds their morals and ethics or how easy they are corrupted
- Moral Alignment= check this:  https://miro.medium.com/v2/resize:fit:1400/1*q0WzYvUsaBdTh0bktmOsmg.png
- Ethics=Explain moral code, how evil or how good they are in relation to X, Y and Z, boundaries, etc.

Subconscious: this section is useful if you're character has trauma or strong beliefs that are more complex than at face value
- Implicit Biases=
- Defense mechanisms=
- Conditioning=root cause, behavior reinforcement, result. 
]

[Section substantiates personality without direct mention unless pertinent:
- Describe your character past here. you don't need to separate it in it's own module if it's something simple and short.
]

[Speech:
- Style=vocabulary, tone, intonation, pacing, sentence structure, voice descriptors, accent.
- Speech pattern references, avoid verbatim, reference only: Greeting: "example". Cornered: "example". Confident: "example". When Safe: "example". Sad: "example". Loath: "example". Surprised: "example". Disgusted: "example". Anger: "example" For more search up "wheel of emotions"
]

Adding Side Characters Module

1
2
3
4
[Relationships:
- Side characeter1(brief physical appearance)=Brief personality and how they are connected to your main character
- Side characeter2=
]

Romantic Stuff and NSFW stuff

  • I usually put these in separate [] module of the usual personality to avoid the LLM turning everything emotional or sexual without it being pertinent
    1
    2
    3
    4
    5
    6
    7
    8
    9
    [Behavior is unadulterated by this section unless pertinent:
    - Primary love languages=
    - Attachment style= Check: https://www.attachmentproject.com/blog/four-attachment-styles/
    - Robert Greene's art of seduction framework of seduction architype=Check this https://www.youtube.com/watch?v=nUgHSxOAlWw
    
    - During sex= Any behavior during sex. usually the llm tends to default to stereotypes if this is not described
    - Kinks=
    - Trait=Any NSFW physical descriptor of your character
    ]
    

Embedded direct prompts

  • You can use direct prompts to talk to the ai in it's own module. the key difference in between this and the usual character section is that in the character section you speak in 3rd person about your character, while in here you speak in 2nd person directly to the ai for example:
    1
    2
    3
    4
    [Ai assistant directives:
    - Remember that CHAR has only one arm and it affects interactions with people and objects.
    - Character is from X fandom, depict accurate X lore.
    ]
    

Extra Bot making info

If your making bots for JLLM, remember the JLLM's total context is 4-7K tokens (counting bot card + custom prompt + your persona)
these are more or less the values of permanent tokens you should aim for for performance sake, however this is personal preference, multiple people have different intervals they like to work on!

Tokens Advice
< 500 Good memory/performance but might feel bland
500 - 1300 Best balanced spot!
1300- 1500 Nice!
1600 - 1700 Cool but need to be smart crafting the prompts!
1800 + No go for multiple characters, they tend to break and have weird behaviors if you do so, be very careful crafting the prompts! It should work ok for single characters tho!
1900+ Okay ish for single characters, might forget something here and there
2000+ DON'T DO IT, high chance for breaking the bot/forgetting things
Edit
Pub: 09 Apr 2025 23:12 UTC
Views: 342